0532

NEW ATTENUATION RELATIONS FOR PEAK GROUND ACCELERATION AND

VELOCITYCONSIDERING EFFECTS OF FAULT TYPE AND SITE CONDITION
Hongjun SI1 And Saburoh MIDORIKAWA2

SUMMARY
New attenuation relationships for peak ground acceleration on soil ground and for peak ground
velocity on stiff ground are developed based on the regression analysis of strong ground motion
recordings from 21 Japan earthquakes. In the analysis, 856 data for peak ground acceleration and
394 data for peak ground velocity are used, including those recorded in the near-source area. In
the study, the earthquakes are classified into three groups, i.e., crustal, inter-plate and intra-plate
earthquakes, according to the fault type. The site effects are evaluated quantitatively, and the path
effects are evaluated by using two types of distance measurement, such as fault distance and
equivalent hypocentral distance. The new attenuation relationships show that an intra-plate
earthquake generates stronger acceleration and velocity than inter-plate and crustal events. The
results also show that the earthquake with deeper focal depth generates stronger ground motion.
INTRODUCTION
The prediction of ground motion from a large scenario earthquake is of fundamental importance to earthquake
engineering. As an empirical method for the prediction, attenuation relations are frequently employed.
Although many attenuation relations have been proposed based on the regression analysis of strong motion
recordings, there remain several problems. First, as most of Japan earthquakes are located in the ocean area, the
near-source recordings are rarely obtained. Few strong motion data in the near-source area are included in the
analysis. Second, it is difficult to express the characteristics of the individual earthquakes in the relation because
the factors used in the relation are limited for simplifying the attenuation model, e.g., magnitude for expressing
source effects. Recently, it is suggested that the fault type and the focal depth should be taken into account for
the source characteristics (McGarr, 1984; Youngs et al., 1997). Third, for the path effects, the appropriate
definition of the distance in the near-source area is in controversy (Campbell, 1985). Fourth, for the site effects,
it is also suggested that the qualitative evaluation using soil types is not sufficient (Fukushima and Tanaka, 1990;
Midorikawa et al., 1994).
In this study, we propose new attenuation relations based on the regression analysis, in which the effects of
source, path and site condition on ground motion are considered more precisely. For the purpose, a strong
motion database is compiled from the recent Japan earthquakes. The database is featured by the following
characteristics: a) a number of near-source recordings is included; b) the magnitude of the earthquakes is widely
distributed in a range of 5.8 to 8.3; c) the earthquakes have various fault types; d) the focal depths range from 6
km to 120 km. In the regression model, in addition to the closest distance to the fault plane, the equivalent
hypocentral distance (Ohno et al., 1993), a recently proposed distance measurement is also adopted. For
evaluating the source effects correctly, we added the fault type and the focal depth as additional factors in the
regression model. In the analysis, the site effects are quantitatively evaluated. Finally, attenuation relations for
peak ground acceleration (PGA) on soil ground, peak ground velocity (PGV) on stiff ground are developed in
this study.

1
2

Kozo Keikaku Engineering Inc., Tokyo, Japan Email: shj@kke.co.jp

Department of Built Environment, Tokyo Institute of Technology, Tokyo, Japan Email: smidorik@enveng.titech.ac.jp

DATABASE
The data from 21 Japan earthquakes occurred in 1968 to 1997 are compiled in the database as shown in Table 1.
Figure 1 shows the distribution in moment magnitude and the focal depth. Here, we define the focal depth as an
average depth of the fault plane. The focal depths in our database are almost less than 40 km, except one is just
over 70 km, and two are deeper than 100 km. Figure 2 shows the histogram of the fault type. The fault type is
classified into 3 categories, that are, crustal, inter-plate and intra-plate earthquakes.
In this study, we use the closest distance to the fault plane (refer to as fault distance) and the equivalent
hypocentral distance (refer to as EHD) to describe the source-to-site distance. The former is simply defined, and
has been often employed in the previous studies. The latter has physical basis, inducing the point source model
to be applicable in the near-source area (Ohno et al., 1993). However, as the distribution of displacement over
the fault plane is needed for computing the EHD, it is not easy to be routinely obtained.
In the database, there are 1137 PGA data and 543 PGV data. All the data are recorded at free field sites or small
buildings where soil-structure interaction effects are negligible. As the data are observed with different types of
the instrument, the data are instrumentally corrected with the filter illustrated in Fig.3. In the figure, the corner
frequency is given as follows according to the noise level of the data: (a) F1=0.15 Hz F2=0.08 Hz; (b) F1=0.20
Hz F2=0.10 Hz; or (c) F1=0.33 Hz F2=0.15 Hz. The data obviously affected by soil liquefaction, such as the
recording at Port Island during 1995 Hyogo-ken Nanbu, Japan earthquake, are removed from the database.
In the analysis, the strong motion data satisfied the following criteria are adopted: distances less than a) 300 km
when magnitude greater than 7, b) 200 km when magnitude between 6.6 and 7, c) 150km when magnitude
between 6.3 and 6.5, and d) 100 km when magnitude under 6.3. Finally, the data used in the analysis are 856
records for PGA, and 394 records for PGV. The peak ground motion is defined as the larger one of the two
horizontal components. In Fig.4, distributions in moment magnitude and fault distance of the data set for PGA
and PGV are illustrated. In the figure, it can be confirmed that a number of near-source recordings is included.
To evaluate the site effects on ground motion, we compiled the soil profiles at the observation stations. First, we
classified soil characteristics into two categories, rock and soil, for all the observation stations. The definition of
rock and soil is after Joyner and Boore (1981). For the stations with more details of soil profile, we calculate the
average shear wave velocity from surface to a depth of 30m [refer to as AVS30 hereafter] at the site. The
AVS30 is computed by dividing the depth by the summation of the travel times of soil layers.
REGRESSION ANALYSIS
Regression Model
Two regression models are used in the analysis. One is for fault distance shown in Eq. (1), another one is for
EHD shown in Eq. (2).
log A = b - log ( X + c ) - k X

(1)

log A = b - log Xeq - k Xeq

(2)

where, A is peak ground motion, X is fault distance in km, and Xeq is EHD in km. The first term in Eqs. (1) (2),
i.e., the coefficient b is an offset factor for each earthquake. The second term shows geometrical attenuation, and
the third term shows anelastic attenuation. For the coefficient k, we fix at 0.003 for PGA and 0.002 for PGV.
For the case of fault distance, coefficient c is introduced accounting for the saturation of the amplitude of strong
motion in the near-source area. The coefficient c is given as a function of magnitude;
(3)

c = c1 10c2 M w
Evaluation of Site Effects on Strong Motion

Recent studies (Campbell, 1991; Midorikawa et al., 1994) indicated that the average amplitude of PGA at soil
sites is about 1.4 times greater than that at rock sites. In the light of these results, we convert the PGA at a rock
site to that at a soil site by multiplying the value with a factor of 1.4. We develop the attenuation relation for
PGA on soil ground because most of the data used in this study are observed at soil sites.

0532

For PGV, Midorikawa et al. (1994) proposed a method for the evaluation of site effects. As shown in Eq. (4),
the site amplification factor R is calculated with AVS30. The PGV at stiff site [Vcor] is calculated by dividing the
observed PGV [Vorg] by R as shown in Eq. (5). The stiff site means that the site amplification factor R shown in
Eq. (4) is 1, i.e., the AVS30 is about 600 m/s. We propose the attenuation relation for PGV on stiff ground.
log R = 1.83 0.66 log AVS30

(4)

Vcor = Vorg / R

(5)

Method
In developing the attenuation relations, we adopted a two-stage regression method. In the first stage, regression
models shown in Eqs. (1) and (3) or Eq. (2) are fitted to the data from each earthquake, gaining the value of
coefficient b. For stronger reflection of the near-source records, we weighted the data closer than 25 km by a
factor of 8, 20 50 km by 4, 50 100 km by 2, and farther than 100 km by 1. For the case of fault distance, the
value of c is difficult to be determined for all the events because of limited number of the near-source recordings.
Here, at first we determined the value of c for the earthquakes with sufficient near-source data, then develop a
regression equation of c as a function of magnitude. Substituting the function to the regression model, the
coefficient b can be obtained. In the second stage, selecting magnitude, the fault type and the focal depth as the
parameters, the regression equation of b is precisely derived.
RESULTS
Analysis without Constraints
Coefficient c in Eq. (1) for PGA and PGV is obtained in the regression analysis of the near-source data from 3
and 4 earthquakes, respectively. The value of coefficient c for each earthquake is plotted versus moment
magnitude in Figs. 5 and 6. In the diagrams, the data of coefficient c obtained from the sufficient near-source
data are plotted with circles. The data of coefficient c from less near-source data, and the datum derived from
the 1985, Chile earthquake (Midorikawa, 1991) are plotted with triangles. In the regression analysis of the data
with model shown in Eq. (3), we derived the relation of the coefficient c and magnitude, as shown in Eq. (6) and
(7), for PGA and PGV, respectively.
c=0.0055 100.50Mw (for PGA)

(6)

c=0.0028 100.50Mw (for PGV)

(7)

In Figs. 5 and 6, the results by Fukushima and Tanaka (1992), and by Nozu et al. (1997) are also plotted. The
results are consistent with those in this study.
By fitting the data from each earthquake to the equation (1) and (6) or (7), the value of coefficient b is obtained.
As an example of the fitting, Figures 7 show the data of PGA and PGV from the 1995 Hyogo-ken Nanbu, Japan
earthquake, and the fitting curve with the regression model. Figures 8 show the relation between coefficient b
and moment magnitude. From the figure, there is strong correlation between coefficient b and magnitude. In the
figure, however, with the same magnitude, the coefficient b has different values. This implied that, in addition to
the magnitude, there are still other factors that affect strong ground motion. Referring to the previous studies,
the focal depth and the fault type are considered as the additional factors. In order to obtain a precise equation
about the coefficient b, a model shown as Eq. (8) is employed.
(8)

b = aM w + hD + d i S i + e +

where, D is focal depth in km, Si is fault type, is the standard

deviation. a, h, d and e is the regression coefficient. Si is a dummy variable, with the value of 1 for each fault
type as described in the previous part and 0 for the others. We weighted the data in the analysis according to the
number of recordings for each earthquake. The weighting factor is set as 3 for the earthquake with mark A in
Table 1, 2 for B and 1 for C.
The coefficients in Eq. (8) derived from the regression analysis are shown in Table 2. For the case of EHD, the
results are listed in Table 3.

0532

Analysis with Constraints

In our database, it is remarkable that there is correlation among the parameters in the regression model. For
example, the earthquake with larger magnitude has deeper focal depth. Because of this, coefficients a and h may
not be correctly determined. Comparing with the results shown in Table 3 for EHD, the results shown in Table 2
for fault distance show larger value of coefficient a for Mw and smaller value of the coefficient h for focal depth.
This is considered to be the influence of the trade-off.
At sites far from the earthquake source, the predicted strong motions by the attenuation relations should be the
same in spite of the difference of distance measurement. For this reason, in the attenuation relations with
different distance measurement, the values of coefficient a for magnitude are likely to be the same. Here, we
assign this to a constraint for the regression analysis. In recent studies, it is indicated that the PGA at the nearsource sites should be almost the same in spite of the difference of magnitude. Thus, We adopted an additional
constraint in the analysis, that is, the predicted PGAs are the same at the sites where the distance is zero.
With these constraints, by performing the regression analysis again, we derived the new results, as shown in
Tables 4 and 5. Substituting the coefficients into the attenuation models, we obtained the new attenuation
relations for PGA and PGV. The log-scaled standard deviations of the new attenuation relations are shown in
Table 6. The standard deviations in this study are less than 0.25 for the distances less than 100 km. It suggests
that the new attenuation relations fitted the data very well.
Predicted Strong Ground Motions
In Figs.9, the attenuation curves are illustrated for the inter-plate earthquakes, with focal depth of 5 km,
magnitudes of 6.0 to 8.0, respectively. In the figures, the attenuation curves for PGA show small magnitude
dependencies, but those for PGV show stronger one. This is consistent with the results in the previous studies.
In Figs.10, the attenuation curves are illustrated for the intra-plate earthquakes with magnitude of 7. The focal
depths are set to 30 km and 100 km. In the figure, the predicted PGA and PGV for the focal depth of 100 km is
about 2 and 1.8 times greater than those for the focal depth of 30 km, respectively. In Figs.11, the attenuation
curves are illustrated for the earthquakes with different fault types. The magnitude is 7, and the focal depth is 20
km. In the figure, the predicted PGA and PGV for the intra-plate event is about 1.7 and 1.4 times greater than
those for the crustal and inter-plate events, respectively.
In Figs.12 and 13, we compare the attenuation relations proposed in this study and those in the previous studies
(Boore et al., 1997; Campbell, 1997; Youngs et al., 1997; Fukushima and Tanaka, 1992; Annaka et al., 1997;
Ohta and Ohno, 1996; Joyner and Boore, 1982; and Kawashima et al., 1985). The comparisons are carried out
for the case of fault distance and EHD, respectively. The predicted curves by the relations in this study are
calculated for the earthquake that Mw is 7, the focal depth is 5 km, and the fault type is crustal earthquake. In the
figure, the results in this study are consistent with those proposed by the previous studies.
CONCLUSIONS
New attenuation relationships for peak ground acceleration on soil ground and for peak ground velocity on stiff
ground are developed based on the regression analysis of strong ground motion recordings from 21 Japan
earthquakes. In the analysis, 856 data for peak ground acceleration and 394 data for peak ground velocity are
used, including those recorded in the near-source area. In the study, the earthquakes are classified into three
groups, i.e., crustal, inter-plate and intra-plate earthquakes, according to the fault type. The site effects are
evaluated quantitatively, and the path effects are evaluated by using two types of distance measurement, such as
fault distance and equivalent hypocentral distance. The new attenuation relationships show that an intra-plate
earthquake generates stronger acceleration and velocity than inter-plate and crustal events. The results also show
that the earthquake with deeper focal depth generates stronger ground motion.
ACKNOWLEDGEMENTS
The authors thank the following organizations for providing the strong motion data and site information:
Building Research Institute, Central Research Institute of Electric Power Industry, Electric Power Development
Co., Fire Research Institute, Fujita Co., Hachinohe Institute of Technology, Hazama Co., Hokkaido Electric
Power Co., Hirosaki University, Hokkaido University, Japan Meteorological Agency, Kajima Co., Kanagawa
Prefecture Government, Kanagawa University, Kansai Electric Power Co., K-NET, Kumagai Co., Kyoto
University, Mitsubishi Estate Co., Matsumura-Gumi, Mitui Co., National Research Institute For Earth Science

0532

and Disaster Prevention, Nishimatsu Constr., NTT, Obayashi Co., Osaka Gas Co., Port and Harbour Research
Institute, Public Work Research Institute, Railway Technical Research Institute, Sato Kogyo Co., Shimizu Co.,
Takenaka Co., Committee of Earthquake Observation and Research in the Kansai Area, Tobishima Co., Tokyo
Gas Co., Tokyo Metropolitan Government, Tokyo Electric Power Co., Tokyo Institute of Technology, Tokyu
Constr., University of Tokyo, and Waseda University.
REFERENCES
Annaka, M., F. Yamazaki, and F. Katahira (1997), A proposal of estimation equations for peak ground motion and
response spectral using the data recorded by JMA87 seismograph, Proceedings of the 24th JSCE Earthquake
Engineering Symposium, pp. 161-164 (In Japanese)
Boore, D.M, W.B. Joyner, and T.E. Fumal (1997), Equations for estimating horizontal response spectra and peak
acceleration from Western North American earthquakes: a summary of recent work, Seism. Res. Lett., Vol.68, pp.128153.
Campbell, K.W (1985), Strong motion attenuation relations: a ten-year perspective, Earthquake Spectra, Vol.1,
pp.759-804 1985
Campbell, K.W. (1991), An empirical analysis of peak horizontal acceleration for the Loma Prieta, California,
earthquake of 18 October 1989, Bull. Seism. Soc. Am., Vol.81, pp.1938-1858.
Campbell, K.W. (1997), Empirical near-source attenuation relationships for horizontal and vertical components of
peak ground acceleration, peak ground velocity, and pseudo-absolute acceleration response spectra, Seism. Res. Lett.,
Vol. 68, pp.154-179.
Fukushima, Y. and T. Tanaka (1990), A new attenuation relation for peak horizontal acceleration of strong
earthquake ground motion in Japan, Bull. Seism. Soc. Am., Vol.80, pp.757-783.
Fukushima, Y. and T. Tanaka (1992), Revised attenuation relation of peak horizontal acceleration by using a new
data base, Program and Abstracts of the Seism. Soc. Japan, No.2, p.116. (In Japanese)
Kawashima, K., K. Aizawa and K. Takahashi (1985), Attenuation relations for peak ground motion and response
spectral, Tech. Memorandum of Public works Res. Inst., No.166 (In Japanese).
Joyner, W.B. and D.M. Boore (1981), Peak horizontal acceleration and velocity from strong motion records including
records from the 1979 Imperial Valley, California, earthquake, Bull. Seism. Soc. Am., Vol.71, pp.2011-2038.
Joyner, W. B. and D. M. Boore (1982), Prediction of earthquake response spectra, U.S. Geol. Surv. Open-File Rept.,
No.82-977.
McGarr, A. (1984), Scaling of ground motion parameters, state of stress, and focal depth, J. Geophys. Res., Vol.89,
pp.6969-6979.
Midorikawa, S. (1991), Attenuation of the peak ground acceleration and velocity during the 1985 Chile and
Nihonkai-chubu earthquakes, Journal of Struct. Construct. Eng., No.422, pp.37-44 (in Japanese, with English
abstract).
Midorikawa, S., M. Matsuoka and K. Sakugawa (1994), Site effects on strong-motion records during the 1987 Chibaken-toho-oki Japan earthquake, The 9th Japan Earthquake Engineering Symposium, Vol.3, pp. 85-90.
Molas, G.L. and F. Yamazaki (1995), Attenuation of earthquake ground motion in Japan including deep focus
events, Bull. Seism. Soc. Am., Vol.85, pp.295-319.
Nozu, A., T. Uwabe, Y. Sato, K. Ichii, and S. Iai (1997), Attenuation relations for peak ground motions on base layer
in Japan, Proceedings of Earthquake Criteria Workshop -Recent Developments in Seismic Hazard and Risk
Assessments for Port, Harbor, and Offshore Structures-, pp.118-135.
Ohno, S., T. Ohta, T. Ikeura, and M. Takemura (1993), Revision of attenuation formula considering the effect of fault
size to evaluate strong motion spectra in near field, Tectonophysics, Vol.218, pp.69-81.
Ohta, T. and S. Ohno (1996), Average characteristics of earthquake ground motions in and around focal region based
on strong motion records, Journal of Struct. Construct. Eng., No.479, pp.31-40 (in Japanese).
Tanaka, T., and Y. Fukushima (1987), Attenuation relations for peak amplitudes of earthquake ground motions,
Proc. 15th Erthq. Ground motion Symp., Architectural Institute of Japan, pp.7-16 (in Japanese).
Youngs, R.R., S.J. Chiou, W.J. Silva, and J.R. Humphrey (1997), Strong ground motion attenuation relationships for
subduction zone earthquakes, Seism. Res. Lett., Vol.68, pp.58-73.

0532

Table 1. The list of the data compiled in the database

NO.

Earthquake

Number of recordings

Depth

Peak acceleration

Fault Type

Peak velocity

Weight

1 Off Tokachi

1968.05.16

8.2

15

10

10 Inter-plate

2 Off Nemuro Pen.

1973.06.17

7.8

25

4 Inter-plate

3 Near Izu Oshima

1978.01.14

6.6

4 Off Miyagi Pref.

1978.06.12

7.6

37

5 East off Izu Pen.

1980.06.29

6.5

6 Off Urakawa

1982.03.21

6.9

7 Nihonkai-Chubu

1983.05.26

7.8

21

17 Inter-plate

8 Off Hyuganada

1984.08.07

6.9

30

8 Intra-plate

9 Central Iwate Pref.

1987.01.09

6.6

73

10

5 Intra-plate

10 Northern Hidaka Mt.

1987.01.14

6.8

120

16

11 East off Chiba Pref.

1987.12.17

6.7

30

173

12 Off Kushiro

1993.01.15

7.6

105

51

13 Off Noto Pen.

1993.02.07

6.3

15

21

14 Southwest off Hokkaido

1993.07.12

7.7

10

52

18 Inter-plate

15 East off Hokkaido

1994.10.04

8.3

35

41

17 Intra-plate

16 Far off Sanriku

1994.12.28

7.7

35

83

30 Inter-plate

17 Hyogo-ken Nanbu

1995.01.17

6.9

10

85

47 Crustal

18 Off Hyuganada

1996.10.19

6.7

25

106

67 Inter-plate

19 Northwestern Kagoshima Pref.

1997.03.26

6.1

121

68 Crustal

20 Northwestern Kagoshima Pref.

1997.05.13

6.0

121

64 Crustal

21 Northern Yamaguchi Pref.

1997.06.25

5.8

10

152

59 Crustal

10

9
F req ue nc y

50

13

10 Inter-plate

19

16 Crustal

25

19

9 Crustal

C
C

9 Intra-plate

47 Crustal

21 Intra-plate

5 Crustal

C
B

C rustal

100

12 Crustal

Inter- p late

6
7

Intra- p late
5

Amplitude

150
Focal depth km

Mw

Date

2
0

7
8
Moment magnitude

F2 F1

Fig.1. Focal depth vs. Mw

10 20
Frequency Hz

Fig.2. Histogram of the fault types

Fig.3. Band-pass filter used for

instrument correction

Mw
Nozu et al. (1997)

100

Coefficient c

Fukushima and Tanaka (1992)

CHILE(1985)

This study
HYOGO-KEN NANBU (1995)

10

EAST OFF IZU PEN. (1980)

VEL.

ACC.
0

100

200

300 0

100

200

NORTHERN YAMAGUCHI PREF. (1997)

ACC.

300

Fault distance km

Mw
Fig.5. Coefficient c for peak ground
acceleration

Fig.4. Fault distance vs. Mw

0532

100

Coefficient c

Nozu et al. (1997)

CHILE (1985)

This study
NIHON-KAI
CHUBU (1983)

10

OFF MIYAGI
PREF. (1978)
HYOGO-KEN
NANBU (1995)

102
P .G.V. cm/s

P .G.A. cm/s 2

103

102

1995.01.17

1995.01.17
1

EAST OFF IZU PEN. (1980)

10 0
10

10 0
10
102
Fault distance k m

101

VEL.
1

d
Crustal Inter-plate Intra-plate
Peak ground acceleration
0.0044
0.00
-0.04
0.17
Peak ground velocity
0.0037
0.00
0.01
0.16
h

Coefficient

Table 2 The results for fault distance

(without constraints)

0.55

102

Fig.7. Examples of fitting the data with the regression model

Fig.6. Coefficient c for peak ground velocity

0.53

101

Mw

101

3
4

0.38
ACCELERAT ION

-1.10

VELOCIT Y

8
9 5
6
Moment magnitude

Fig.8. Coefficient b vs. moment magnitude

Table 3 The results for EHD (without constraints)

0.58
0.60

d
Crustal Inter-plate Intra-plate
Peak ground acceleration
0.0039
0.00
0.01
0.18
Peak ground velocity
0.0032
0.00
0.03
0.13
h

103

M w = 8.0

0.12
-1.39

Table 4 The results for fault distance (with constraints)

a
0.50
0.58

d
h
Crustal Inter-plate Intra-plate
Peak ground acceleration
0.0036
0.00
0.09
0.28
Peak ground velocity
0.0031
0.00
0.06
0.16

P.G.A. cm/s2

7.5
7.0

102

6.5
6.0

ACC.
Depth = 5 k m
Inter-plate

e
0.60

101
102

-1.25

M w = 8.0

a
0.50
0.58

d
Crustal Inter-plate Intra-plate
Peak ground acceleration
0.0043
0.00
0.01
0.22
Peak ground velocity
0.0038
0.00
-0.02
0.12
h

P.G.V. cm/s

Table 5 The results for EHD (with constraints)

e
0.61

Equivalent hypocentral distance

P.G.A.
P.G.V.
All data
0.28
0.23
D 100km
0.24
0.22
D = distance

7.0
6.5

-1.29

Table 6. The standard error for the proposed attenuation

relations
Standard error

7.5

101

Fault distance
P.G.A.
P.G.V.
0.27
0.23
0.25
0.23

VEL.
Depth = 5 k m
Inter-plate
100 0
10

101
Fault dista nce k m

6.0

102

Fig.9. Predicted attenuation curves

0532

103

103
Depth = 30 km
Depth = 100 km

Inter-plate

P.G.A. cm/s2

P.G.A. cm/s2

Intra-plate

102
Crustal

102

ACC.
M w = 7.0
Int ra-p lat e

ACC.
M w = 7.0
Depth = 20 km

101

101
102

Depth = 30 km
Depth = 100 km

P.G.V. cm/s

P.G.V. cm/s

Intra-plate
Crustal
1

10

Inter-plate
VEL.
M w = 7.0
Depth = 20 km

100

101

101

VEL.
M w = 7.0
Int ra-p lat e

100
101

102
Fault distance k m

Fig.10. The effects of focal depth on strong motion

102
Fault distance k m

103

Fig.11. The effects of fault type on strong motion

103

P .G .A . c m /s 2

ACC .
M w = 7.0

102

ACC .
M w = 7 .0

C ru s t al
D ep th = 5 k m

101 0
10

C ru s t al
D ep t h = 5 k m
T h is s tu d y
A n n ak a et a l.(1 9 9 7 ), So il
O h ta an d O h n o (1 9 9 6 ), St iff s o il

T h is s tu d y
B o o re et a l.(1 9 9 7 ), V 30 = 2 0 0 m /s
C am p b ell(1 9 9 7 )
Y o u n gs et a l.(1 9 9 7 ), So il
F u k u s h im a an d T an ak a(1 9 9 2 )

101

102

F a ult d is ta nc e k m

101

102

E q uiva le nt H yp o c e ntra l D is ta nc e k m

Fig.12. Comparison of the attenuation relations for peak ground acceleration

102

P.G.V. cm/s

VEL.
Mw = 7.0

VEL.
Mw = 7.0

101
Crustal
Depth = 5 km
This study, V30 = 600 m/s
Annaka et al(1997), Rock
Joyner and Boore(1982), Rock
Kawashima et al(1985), Soil typ e 1

100 0
10

101

Crustal
Depth = 5 km
This study , V30 = 600 m/s
Annaka et al(1997), Rock
Ohta and Ohno(1996), Stiff soil

102

Fault distance k m

101

102

Equivalent Hypocentral Distance k m

Fig.13. Comparison of the attenuation relations for peak ground velocity

0532